wtf is tradingview-mcp?

Tradingview MCP is a small server that lets Claude Code, or any other MCP-compatible AI assistant, control a TradingView Desktop application that is already running on your computer. MCP stands for Model Context Protocol, a standard way for AI tools to call external tools. Instead of scraping TradingView's website or getting around any paywall, this project connects to the desktop app you already log into and pay for, using the Chrome DevTools Protocol, the same debugging interface built into Electron apps like Slack, VS Code, and Discord. Once connected, an AI assistant can read the current chart, including the symbol, timeframe, visible indicators, and last price, and it can change the symbol or timeframe on your behalf. It can also fetch up to 5,000 recent OHLCV bars, meaning open, high, low, close, and volume price data, directly from the active chart. All of this happens locally: no chart data is sent anywhere else or stored by the tool itself. The project deliberately keeps its feature set small, with a handful of well defined, typed tools rather than a large kitchen sink set of features. It includes a doctor command that checks whether the connection to TradingView Desktop is working and reports what is wrong if it is not. Setup involves quitting TradingView Desktop, relaunching it with a debug port flag enabled, and then adding the tool to your Claude Code configuration. An optional local Redis cache can be used to reduce repeated calls to the chart, and the tool still works without Redis running, just without that caching benefit. The code is written in TypeScript with strict type checking, split into separate layers for the connection, the chart tools, and the server itself, and it ships with tests. This is aimed at developers who already use TradingView Desktop and want an AI assistant to read chart state or pull price data as part of a workflow, such as asking an assistant to switch symbols or summarize recent price movement. It is not affiliated with TradingView Inc. and requires an existing TradingView subscription and a locally running desktop app to work.